Improving Brain Age Prediction with Low-Rank Tensor Fusion
Author Information
Author(s): Liu Xia, Zheng Guowei, Beheshti Iman, Ji Shanling, Gou Zhinan, Cui Wenkuo
Primary Institution: Hebei University of Economics and Businesses
Hypothesis
Can low-rank tensor fusion enhance the accuracy of brain age estimation using multimodal neuroimaging data?
Conclusion
The low-rank tensor fusion algorithm effectively integrates multimodal data into deep learning frameworks for estimating brain age.
Supporting Evidence
- The model achieved an R2 value of 0.98 and a mean absolute error of 2.25 years on the training set.
- On the test set, the model obtained a mean absolute error of 4.20 years and an R2 of 0.93.
- Low-rank tensor fusion outperformed traditional feature fusion methods in predicting brain age.
Takeaway
This study shows that combining different brain imaging techniques can help us better understand how our brains age.
Methodology
The study developed a low-rank tensor fusion technique to integrate structural MRI, diffusion tensor imaging, and magnetoencephalography data for brain age estimation.
Potential Biases
The model failed to converge when using MEG data alone, indicating potential issues with model training.
Limitations
The study's sample size is limited, and the model's performance on independent datasets is not validated.
Participant Demographics
521 healthy controls, aged 18–88, with 270 males and 251 females.
Statistical Information
P-Value
<0.001
Statistical Significance
p<0.001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website